Spatial Audio Reconstruction for VR Applications using a Combined Method based on SIRR and RSAO approaches

In order to recreate a sound field as realistic and immersive as the original setting, it is important to preserve the properties of the room acoustics. The room acoustics are usually captured by room impulse responses (RIRs) which can be employed to generate sound that could lead to the same audio perception. In this paper, we compare and combine two main parametric approaches, Spatial Impulse Response Rendering (SIRR) and Reverberant Spatial Audio Object (RSAO) to encode and render the RIRs. We first discuss that SIRR synthesises the early reflections more precisely whereas RSAO is a better approach to render the reverberation. For our proposed combined method, each RIR is divided into three parts: The direct sound, early reflections, and late reverberation, as in RSAO. To estimate the boundary time between the early reflections and late reverberation, we employ the diffuseness factor as in SIRR and set the mixing time as the time when the diffuseness factor reaches a threshold. Then the early part is analysed and synthesised by SIRR while the reverberant tail is encoded and rendered using RSAO. We show that in general, the combined method benefits the positive aspects of the two baseline methods.